Numerical linear algebra on emerging architectures: The PLASMA and MAGMA projects

By Emmanuel Agullo, J. Demmel, Jack J. Dongarra, Bilel Hadri, Jakub Kurzak, J. Langou, Hatem Ltaief, Piotr Luszczek, S. Tomov. 2009

486 citations recorded by the source corpus.

The emergence and continuing use of multi-core architectures and graphics processing units require changes in the existing software and sometimes even a redesign of the established algorithms in order to take advantage of now prevailing parallelism. Parallel Linear Algebra for Scalable Multi-core Architectures (PLASMA) and Matrix Algebra on GPU and Multics Architectures (MAGMA) are two projects that aims to achieve high performance and portability across a wide range of multi-core architectures and hybrid systems respectively. We present in this document a comparative study of PLASMA's performance against established linear algebra packages and some preliminary results of MAGMA on hybrid multi-core and GPU systems.

Subfields: Computational Theory and Mathematics, Computer Networks and Communications, Hardware and Architecture

View Paper: https://www.semanticscholar.org/paper/917fd12162c12c0fd2cb6409de1dd438531c553a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Computer Networks and Communications

In Hardware and Architecture

Citations in the corpus

BLIS: A Framework for Rapidly Instantiating BLAS Functionality (2015, unknown): “More recently, other approaches to parallelism have been explored by the developers to unleash GPUs for parallel computing, the PLASMA project [Agullo et al. 2009], which explores parallelism via task scheduling of what they call “tiled” algorithms, and Quark [Agullo et al. 2011], which more…”
High-order finite-element seismic wave propagation modeling with MPI on a large GPU cluster (2010, methodology): “Dense BLAS level 3 operations, such as SGEMM and DGEMM, can be implemented to achieve close to peak arithmetic throughput, using well-known tile-based techniques [39,40].”
The analysis of a plane wave pseudopotential density functional theory code on a GPU machine (2013, unknown): “0 and ELPA (git version) [17,18].”
GPU-accelerated preconditioned iterative linear solvers (2013, methodology): “For dense linear algebra computations, the Matrix Algebra for GPU and Multicore Architectures (MAGMA) project [2] using hybrid multicore-multiGPU system aims to develop a dense linear algebra library similar to LAPACK.”
DAGuE: A Generic Distributed DAG Engine for High Performance Computing (2011, background): “Recently, several projects [4], [5], [6], [7], [8], mostly in the ﬁeld of Linear Algebra, have proposed to use of DAGs as an approach to tackle the challenges of harnessing the power of multi-core and hybrid platforms.”
PLASMA (2019, methodology): “Initial work on tile algorithms was published in [Agullo et al. 2009; Buttari et al. 2009], and a recent overview for the development of tile algorithms can be found in [Abdelfattah et al.”